Method for Constructing Training Data Set in Intrusion Detection System
Chen Gong-he · Computer Engineering and Applications Journal · 2006
Taking the example of designing classifier in intrusion detection system,this paper studies the method for constructing training data set in pattern classification.A constructing method is proposed fitting for asymmetric distributing and large data set.First,through retaining the boundary and discarding most of the interior training samples,it selects important samples from classes containing more samples.Then constructs virtual samples for the class that contains lesser samples.